JPH0743252B2 - Wiring pattern inspection device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a wiring pattern inspecting apparatus for inspecting a wiring pattern on a printed circuit board, a photomask or the like.

[0002]

2. Description of the Related Art Conventionally, visual inspection by humans has been used to inspect defective printed circuit boards. However, as products become smaller and lighter, the wiring patterns of printed circuit boards are becoming finer and more complex. Under such circumstances, it is becoming difficult for a person to keep a very fine wiring pattern for a long time while maintaining high inspection accuracy, and automation of inspection is strongly desired.

An example of a wiring pattern defect detecting device is disclosed in Japanese Patent Laid-Open No. 04-73759.
This defect detection method is a feature extraction method based on the design rule method, and a process of removing one pixel from the background side of the wiring pattern using a binary image is repeated n times to obtain a skeleton image. The feature points are extracted, and one or more neighboring features extracted in advance on a non-defective board are compared and determined with the feature information added as supplementary information, and high-speed search can be performed based on the supplementary information. It is described that the alignment can be facilitated by comparing the relative positions of.

FIG. 31 is a flow chart showing the operation of a conventional wiring pattern inspection apparatus. The operation will be described below with reference to FIG.

(A) First, the relative values of the coordinates of the inspection data from the feature extraction means and all the learning data indicated by the supplementary information of the previous learning data from the characteristic information storage means are calculated.

(B) Next, the relative value of the coordinates and the detection information of the inspection data from the feature extraction means and all the learning data indicated by the supplementary information of the previous learning data from the characteristic information storage means are compared and determined.

(C) If the comparison results are in agreement, it is the characteristic information registered in the learning data and is not a defect.

(D) If they do not match as a result of comparison, the defect is notified or registered. The above processing is repeated until there is no more characteristic information data.

[0009]

As described above, the method for detecting the defect of the wiring pattern in the binarized image based on the design rule method has been described. This method can be said to be a promising method because it is possible to reliably detect defects.

However, the wiring pattern of the printed circuit board is
It is well known that variations in the line width and the coordinates of characteristic points occur due to deformation (including expansion and contraction) and distortion of the substrate in each manufacturing process. Similarly, when inspecting, the problem of alignment with reference data cannot be overlooked. Further, when there are a plurality of candidates for comparison determination or when the feature extraction is performed by the thinning process, the generation of the skeleton image becomes unstable due to the variation of the etching process, which causes a false alarm.

In view of the above problems, the present invention provides a stable wiring pattern inspection apparatus which absorbs expansion / contraction and distortion of a wiring pattern, a positioning error, and the like and has few false reports.

[0012]

In order to solve the above-mentioned problems, the technical solution of the present invention is, firstly, image input means for photoelectrically converting a wiring pattern formed on a printed circuit board,
Binarization means for converting the grayscale image from the image input means into a binary image, feature extraction means for extracting the features of the wiring pattern from the binary image from the binarization means, and the feature extraction means From the characteristic information storage means, the characteristic information storage means for temporarily storing the characteristic information of the wiring pattern, the reference information storage means for preliminarily extracting the characteristic information of the wiring pattern from the characteristic extraction means from the non-defective substrate and storing it as the reference data, and the characteristic information storage means. The characteristic information of the wiring pattern of the inspected substrate and the characteristic information of the non-defective substrate from the reference information storage means are arbitrarily determined around the characteristic point tn of interest extracted from the inspected substrate as the first determination process. It is determined whether or not the feature point rn from the non-defective substrate corresponds to the first determination area, and if the corresponding feature point does not exist, it is registered as a defect, and there are a plurality of corresponding feature points. Feature point t as a second determination process
One or more peripheral feature points t that have already been associated with n
A second determination region arbitrarily determined with a distance sl from the peripheral feature point rn-i already associated with the peripheral feature point tn-i of the non-defective substrate as a center. The comparison determination means determines whether there is a feature point rn corresponding to the target feature point tn, and if it does not exist, registers it as a defect and detects only a true defect.

Secondly, in the comparison / determination means, the feature point rn from the non-defective substrate corresponds to an arbitrarily determined first determination region centered on the feature point tn of interest extracted from the inspected substrate as the first determination process. If there is no corresponding feature point, it is registered as a defect, and if there are a plurality of corresponding candidate points, the feature point tn of interest as the second determination process.
And the distance to one or more peripheral feature points, and for each candidate point of the non-defective substrate, an arbitrary second point centered on the distance from the peripheral feature point
It is determined whether or not peripheral feature points exist in the determination area, a score is given each time a feature point exists, and the candidate point with the highest score is selected as the corresponding feature point rn.

Thirdly, in the comparison / determination means, the feature point rn from the non-defective substrate corresponds to an arbitrarily determined first determination region centered on the feature point tn of interest extracted from the inspected substrate as the first determination process. If there is no corresponding feature point, it is registered as a defect, and if there are a plurality of corresponding feature points, the target feature point tn as the second determination process.
And one or more peripheral feature points tn already associated with
A distance sl with respect to −i is obtained, and there is a feature point rn corresponding to a second determination region arbitrarily determined around a point with a distance sl from the peripheral feature point rn-i of the non-defective substrate that has already been associated. If there is no corresponding candidate point and a plurality of corresponding candidate points cannot be specified, the distance between the target feature point tn and one or more peripheral feature points is calculated as the third determination process. For each candidate point on the non-defective board, it is determined whether or not a peripheral feature point exists in a third determination region that is arbitrarily set around the point of distance from the peripheral feature point, and the score is calculated for each feature point. Feature point r that corresponds to the candidate point with the highest score
This is selected as n.

Fourthly, in the reference information storage means, the characteristic information of the wiring pattern from the characteristic extracting means is composed of a series of numbers in the order of detection, coordinates, a code capable of identifying the characteristic, and a series of a plurality of peripheral characteristic points. It is the one.

Fifth, when obtaining a plurality of peripheral feature points to be registered as one of the feature information of the reference information storage means, the region is divided into four with the target feature point as the center, and the feature of the closest distance in each region. The point is selected.

Sixth, in the comparison / determination means, a feature point map obtained by dividing the entire area of the printed circuit board into an arbitrary size is set, and the feature information from the feature extraction means is registered in the corresponding area. The peripheral feature points are searched from the feature point map.

Seventh, in the comparison / determination means, the first feature information from the feature extraction means is registered in the corresponding area in the feature point map obtained by dividing the entire area of the printed circuit board into an arbitrary size, and the divided area is also divided. When registering the second and subsequent feature points existing in the map, a serial number of the feature point is added as map information to the feature information of the reference information storage unit detected one before, and the feature points around the feature point of interest are added. Are searched in a chain from the feature point map and the map information added to the feature information of the reference information storage means.

Eighthly, when the comparison determination means compares and determines the feature point of interest from the feature information storage means, the peripheral feature points are registered as the next search candidates, and after the comparison determination of the feature points of interest is completed, the next search candidate is selected. The following feature points of interest are searched in a chain.

Ninth, the comparison / determination means sequentially corrects the coordinates of the feature points from the reference information storage means with the average value of the deviation amounts of the one or more feature points already determined.

Tenth, in the comparing and judging means, the characteristic extracting means converts the binary image into a skeleton image by thinning processing or the like to extract characteristics of the wiring pattern such as end points and branch points, and the characteristic points of the characteristic points are judged at the time of comparing and judging. The feature identification code is reattached according to the combination of feature points in an arbitrary neighborhood area.

Eleventh, in the comparing and determining means, when the characteristic extracting means converts the binary image into a skeleton image by a thinning process of n times for a limited number of times to extract the characteristics of the wiring pattern such as end points and branch points. In addition, a flag is added to the feature point during the nk-th thinning process and thereafter, and the feature point is excluded when an error is determined during the comparison determination.

[0023]

According to the present invention, firstly, in the comparing and determining means,
It is determined whether the feature point rn from the non-defective substrate corresponds to the arbitrarily determined first determination region centered on the feature point tn of interest extracted from the inspected substrate in the determination process of A distance sl between the feature point tn and one or more peripheral feature points tn-i that have already been associated is obtained, and a point with a distance sl from the peripheral feature point rn-i that has already been associated with the non-defective substrate is centered. A plurality of candidate points can be surely associated with each other by determining whether or not there is a feature point rn corresponding to the arbitrarily determined second determination area.

Secondly, in the comparison / determination means, the feature point rn from the non-defective substrate corresponds to an arbitrarily determined first determination region centered on the feature point tn of interest extracted from the inspected substrate as the first determination process. If there is a plurality of corresponding candidate points, the target feature point tn is determined as the second determination process.
And the distance between the peripheral feature point and one or more peripheral feature points, and whether or not the peripheral feature point exists in a second determination area arbitrarily determined around the point of the distance from the peripheral feature point for each candidate point of the non-defective substrate. The determination is made and a score is given every time there is a feature point, and the candidate point with the highest score is selected as the corresponding feature point rn. Therefore, a plurality of candidate points can be surely dealt with.

Thirdly, if the comparison and determination means cannot determine the first and second determination processes, the third determination process is performed.
As the determination processing of, the distance between the feature point tn of interest and one or more peripheral feature points is obtained, and the feature point exists in the third determination region that is arbitrarily determined around the point of distance from each candidate point of the non-defective substrate. Whether or not to do so, a score is given each time there is a feature point, and the candidate point with the highest score is selected as the corresponding feature point rn, so that a plurality of candidate points can be reliably associated.

Fourthly, the characteristic information of the wiring pattern is made up of a series of numbers in the order of detection, coordinates, a code capable of identifying the characteristic, and a series of a plurality of peripheral characteristic points to facilitate the retrieval of peripheral points. It is a thing.

Fifth, when obtaining a plurality of peripheral feature points to be registered as one of the feature information of the reference information storage means, the region is divided into four with the target feature point at the center and the feature of the closest distance in each region. The points are selected so that the surrounding feature points are not concentrated in one place.

Sixth, in the comparing and determining means, a feature point map obtained by dividing the entire area of the printed circuit board into arbitrary sizes is set, and the feature information from the feature extracting means is registered in the corresponding area, and the feature points of interest are The peripheral feature points are searched from the feature point map so that they can be searched at high speed.

Seventh, in the comparison / determination means, the first feature information from the feature extraction means is registered in the corresponding area in the feature point map obtained by dividing the entire area of the printed circuit board into an arbitrary size, and the divided area is divided. When registering the second and subsequent feature points existing in the map, a serial number of the feature point is added as map information to the feature information of the reference information storage unit detected one before, and the feature points around the feature point of interest are added. Is to be searched in a chained manner from the feature point map and the map information added to the feature information of the reference information storage means, and the memory size of the feature point map is reduced.

Eighthly, when the comparison / determination means compares and determines the feature point of interest from the feature information storage means, peripheral feature points are registered as next search candidates, and after the comparison determination of the feature point of interest is completed, the next search candidate is selected. The next feature point of interest is searched in a chained manner, and the processing is performed sequentially from the one having the strongest relevance, so that the processing can be efficiently performed.

Ninth, in the comparison and determination means, the coordinates of the feature points from the reference information storage means are sequentially corrected by the average value of the deviation amounts of the one or more feature points that have already been determined, and the alignment is performed. Making it easy.

Tenthly, in the comparison / determination means, the feature extraction means converts the binary image into a skeleton image by thinning processing or the like to extract the features of the wiring pattern from the end points or branch points, and at the time of comparison / determination, an arbitrary feature point is selected. The identification code of the feature is reattached according to the combination of the feature points in the vicinity region, and the inspector that absorbs the instability due to the variation of the etching process and does not provide false information is provided.

Eleventh, in the comparison / determination means, when the characteristic extraction means converts the binary image into a skeleton image by a thinning process of a limited number of times n times, and extracts the characteristics of the wiring pattern from the end points or branch points, A flag is attached to the feature point during the thinning process from the n-th time onward, and the feature point is excluded when it is determined to be an error in the comparison determination. Is provided.

[0034]

【Example】

(Embodiment 1) Hereinafter, a first embodiment of the present invention will be described.

FIG. 1 shows a wiring pattern inspection apparatus 1 according to the present invention.
It is a block connection diagram showing an example. In FIG. 1, 1
01 is a printed circuit board, 103 is a diffuse illumination device 104 and C
An image input unit equipped with a photographing device 102 such as a CD camera, 105 a binarizing unit for converting a grayscale image into a binary image, 106 a feature extracting unit for extracting a feature of a wiring pattern from the binary image, and 107 a Reference information storage means for extracting the characteristics of the wiring pattern from the non-defective board in advance and storing it as reference data, 108 is characteristic information storage means for extracting and temporarily storing the characteristics of the board to be inspected, and 109 is a CPU (central processor unit). ) 110 and error information memory 112.

The operation of the wiring pattern inspecting device constructed as above will be described. First, the printed circuit board 101 to be inspected is imaged by the image input means 103 using a CCD camera or the like to obtain a grayscale image. In this embodiment, an example using a CCD line sensor camera as the image input means 103 will be described. The illumination system is not particularly limited, but it is well known that binarization can be easily performed by using a wavelength near 600 nm in which the difference in reflection luminance between the base material (glass epoxy or the like) and the copper pattern is large. Recently, ultra-high brightness LEDs (eg GaAlAs 660nm)
Alternatively, InGaAlP 620 nm) has been commercialized, and it can be said that arranging on a line and irradiating from multiple directions is also an effective illumination method. The obtained grayscale image is binarized 105
Then, it is converted into a binary image at a threshold level previously obtained from a density histogram or the like. Here, the wiring pattern side is "1"
Then, the base material side is binarized to "0". The defect extraction means 106 is not particularly limited, but may be any detection method capable of detecting defects in pixel units. For example, as shown in the conventional example, the pixels of the background of the wiring pattern are removed one by one while the thinning process is repeatedly performed on all the pixels, and the obtained skeleton image is used as the characteristic information of the wiring pattern to identify the coordinates and the identification information of the characteristic points. (End points and branches) are extracted. The reference information storage means 107 stores the characteristic information from the characteristic extraction means 106 on a non-defective substrate in advance as reference data. The characteristic information storage unit 108 temporarily stores the characteristic information from the board to be inspected. The comparison / determination means 109 is composed mainly of a CPU system 110 and an error information storage memory 112 for storing defect information, and is mainly realized by software processing. The characteristic information from the characteristic information storage means 108 and the reference information storage means 107. At the time of comparison and determination, the first determination region, which is arbitrarily determined and is centered around the feature point tn of interest extracted from the inspected substrate in the first determination process, is set as the feature point rn from the non-defective substrate.
And if there is no correspondence, the distance sl between the target feature point tn and one or more peripheral feature points tn-i that have already been associated is obtained in the second determination process, and a non-defective product is obtained. A plurality of candidates exist by setting an arbitrary second determination area around a point having a distance sl from the feature point rn-i that has already been associated with the substrate and determining whether the corresponding feature point rn exists. Only true defects are extracted so that points can be surely associated with each other. The extracted defects are registered in the error information memory 112. Here, the distance means what is expressed by the distance and the direction.

By repeating the above operation and sequentially performing it, the entire surface of the printed circuit board 101 can be inspected. This series of operations is performed in synchronization with an appropriate signal.

Next, the comparison / determination means 109 different from the conventional example.
This will be described in detail with reference to FIGS. 2A to 2D, FIGS. 3A and 3B, and FIG. 2 (a) to (d),
The coordinates of the non-defective board and the board to be inspected are shown, and the comparison / determination means 10
The concept of the first determination processing of No. 9 is described, and it is compared and determined whether the detected target feature point tn (Xn, Yn) corresponds to the feature point rn of the non-defective substrate. In the coordinate system 203 of the reference information, the first determination region 205 having a radius r1 centered on the same coordinates (Xn, Yn) as the feature point tn202 of interest.
It is determined whether or not the characteristic points rn of the non-defective substrate correspond to each other, and it can be determined that the characteristic points rn204 of the non-defective substrate correspond to each other 1: 1 as shown in FIG. FIG. 2C shows a case where the corresponding feature point does not exist, and is registered in the error information memory 112 as a defect. in this case,
If the identification information is different even if the coordinates are the same, it is registered as an error.

FIG. 2D shows the corresponding feature point rn204.
In addition to this, when there are a plurality of feature points of the feature point 206, the first determination process cannot determine and therefore the second determination process is entrusted.

FIG. 3A and FIG. 3B show comparison / determination processing 109.
The second determination process is described below. When a plurality of feature points are present in the first determination region and cannot be determined by the first determination process as shown in FIG. It is something to do. A plurality of feature points 303 (tn-i, tn-) that have already been associated with the detected feature point tn302 of interest.
j), the distances sli304 and slj305 are obtained, and the feature point 3 already associated with the coordinate system 306 of the reference information
The second determination region 311 having a radius r2 is set around each distance previously obtained from 08 (rn-i, rn-j). Therefore, the feature point r corresponding to the second determination area
The determination is made based on whether or not n exists. Figure 3
In (b), although two feature points existed in the determination area in the first determination process, the feature point rn307 is selected as the corresponding feature point in the second determination process. Here, the arbitrarily determined determination area is described as being within the range of the radii r1 and r2, but it may be a rectangular area. Next, a processing flow of the comparison determination processing 109 is shown in FIG. 5 and will be described. As the process a, a plurality of pieces of characteristic information centering on the characteristic point of interest are read from the characteristic information storage means 108 and the reference information storage means 107. As the processing b, the first determination area is set around the same coordinate point of the feature point of interest. As the process c, it is compared and judged whether or not the corresponding characteristic points of the non-defective substrate exist in the arbitrarily set first judgment region. If the corresponding reference feature point exists, the process g is executed and the process is repeated until the process is completed.

If there is no corresponding point, the process f is executed. Further, when there are a plurality of corresponding points, the distances between the target feature point and the plurality of peripheral feature points that have already been associated with each other are obtained as processing d, and the distances from the corresponding peripheral feature points of the non-defective substrate are calculated. The second determination area is obtained centered on.

As the processing e, it is determined whether or not the corresponding feature point of the non-defective substrate exists in the second determination area. If the corresponding feature point exists, the process g is executed as the completion of the association, and if there is no corresponding point, the process f is executed as a defect. The process f registers the defect information in the error information memory.

The process g is for determining whether the determination process is completed, and is, for example, a case where the process is completed for all the feature information in the feature information storage means. If not, return to the starting point.

The above processing is repeated until there is no more characteristic information data from the characteristic information storage means. Processing h
Is a feature point that has not been described in the previous process but has been associated with the collation flag added to the feature information in the reference information storage means. The collation flag is checked, the characteristic information to which the collation flag is not added is registered as an error in the error information memory, and the process ends.

In this embodiment, the characteristic information data from the characteristic information storage means has been described as the characteristic point of interest, but the same result can be obtained even if the reference information data from the reference information storage means is used as the characteristic point of interest. is there.

(Second Embodiment) A second embodiment of the present invention will be described below.

FIGS. 3 (c) and 3 (d) are conceptual diagrams of the second judgment processing in the comparison judgment means 109 showing the second embodiment, and FIG. 5 is a processing flow of the comparison judgment means 109. In the second determination processing of the comparison determination means 109, the distance between the feature point of interest and one or more peripheral feature points is obtained, and the second distance is arbitrarily set around the point of the distance from each candidate point of the non-defective substrate. It is determined whether or not a feature point exists in the determination area, a score is given each time the feature point exists, and the candidate point with the highest score is selected as the corresponding feature point. In the comparison / determination unit 109, the first determination process is (Example 1).
Since it is the same as the above, the description thereof will be omitted, and the second determination process will be described in detail with reference to FIGS.

The second judgment processing is carried out when a plurality of characteristic points are present in the first judgment area and cannot be judged by the first judgment processing as shown in FIG. 2 (d). .
In FIG. 3C and FIG. 3D, the feature point 31 of interest
4 and the peripheral feature points 315 are respectively calculated, and distances 1, 1, 2 and 13 from the respective candidate points 317 and 318 in the coordinate system 316 of the reference information are used as the center of the second distance. The determination area 321 is set, and a score is given according to how many peripheral feature points corresponding to each candidate point exist in the second determination area, and the candidate point with the highest score is set as the corresponding feature point. . In the example of FIG. 3D, the candidate point 3
Since there are three peripheral feature points in the determination region of 17 and two peripheral feature points in the determination region of the candidate point 318, the candidate point 317 having a high score is selected as the corresponding feature point. It is something.

Next, a processing flow of the comparison / determination processing 109 will be described with reference to FIG. As the process a, a plurality of pieces of feature information centering on the feature point of interest are read from the feature information storage means 108 and the reference feature information storage means 107. As the processing b,
The first determination area is set around the same coordinate point of the feature point of interest. As the processing c, it is compared and judged whether or not the corresponding characteristic points of the non-defective substrate exist within the set judgment area. If there is a feature point to be associated, process g is executed and the process is repeated.
If there is no corresponding point, the process f is executed. When there are a plurality of corresponding candidate points, the distance between the target feature point and one or more peripheral feature points is obtained as the second determination processing d, and the respective distances from each candidate point of the non-defective substrate are calculated. The second determination areas are obtained in the center. As the processing e, if the characteristic points are present in the respective set judgment areas, the respective points are given. The process i selects a candidate point having a high score as a corresponding feature point. If the scores are the same, it is assumed that the one having the closer coordinates (distance) of the candidate points is selected, and the process g is executed. The process f registers the defect information in the error information memory 112. The process g is to determine whether the determination process is completed, and for example, the feature information storage unit 10
This is the case, for example, when the processing is completed for all the characteristic information items of No. 8. If not, return to the starting point.

The above processing is repeated until there is no more characteristic information data from the characteristic information storage means 108.
In the process h, the feature points which have not been described in the previous process but have been associated are the feature information in the reference information storage unit to which the collation flag is added, and the final feature is all the features in the reference information storage unit. The collation flag of the information is checked, the characteristic information to which the collation flag is not added is registered as an error in the error information memory, and the process ends.

In the present embodiment, the distances between the feature point of interest and one or more peripheral feature points are obtained when the second determination process is performed, and the respective distances from each candidate point of the non-defective substrate are set to the center. Although the second determination area is obtained in each of the above, the similar processing may be performed by obtaining the distance between the feature point of interest and the peripheral feature points that have already been associated. In this case, the accuracy is improved, but the number of peripheral feature points may be decreased.

(Embodiment 3) A third embodiment of the present invention will be described below. FIG. 4 is a conceptual diagram of the comparison determination means 109 showing the second embodiment, and FIG. 7 is a comparison determination means 109.
Is a processing flow of. In the present embodiment, if the comparison determination means 109 described in (Embodiment 1) cannot specify the first determination processing and the second determination processing, the third determination processing includes the feature point of interest and one or more The distance from the peripheral feature point is obtained, and it is determined whether or not the feature point exists in a third determination region that is arbitrarily defined centering on a point having a distance sl from the candidate point of the non-defective substrate, and each time the feature point exists. A candidate point with the highest score is selected as a corresponding feature point.

In the comparison / determination means 109, the first determination process and the second determination process are the same as those in the first embodiment, so the description thereof will be omitted and the third determination process will be described.

The third determination process is performed by the second determination as shown in FIG.
In the determination process (1), the determination is performed when there are two candidate points 307 and 322 in the second determination region 311 and the corresponding feature points cannot be specified. FIG. 7 shows the processing flow. The defect determination is performed in three stages of the first determination processing 601, the second determination processing 602, and the third determination processing 603 as the flow of the entire processing. As described above, the first determination processing 601 and The second determination process 602 is similar to that of the first embodiment, and therefore its description is omitted.

Process a performs the first determination process and process b performs the second determination process. The process c is a process to be performed when it cannot be specified by the second determination process, and obtains the distances between the feature point of interest and one or more peripheral feature points, and calculates the distances from each candidate point of the non-defective substrate. A third judgment area is obtained centered on. As the process d, if the feature points are present in the respective set determination areas, the scores are given. Processing e
Selects a candidate point with a high score as the corresponding feature point. In addition, when the scores are the same, the coordinate having the closer coordinates of the candidate points is selected, and the process g is executed. The process f registers the defect information in the error information memory 112. The process g is to determine whether the determination process is completed, and is, for example, a case where the process is completed for all the feature information in the feature information storage unit 108. If not, return to the starting point.

The above processing is repeated until there is no more characteristic information data from the characteristic information storage means 108.
In the process h, the feature points which have not been described in the previous process but have been associated are the feature information in the reference information storage unit to which the collation flag is added, and the final feature is all the features in the reference information storage unit. The collation flag of the information is checked, the characteristic information to which the collation flag is not added is registered as an error in the error information memory, and the process ends.

(Fourth Embodiment) The fourth embodiment of the present invention will be described below. FIG. 8 is a format of the characteristic information of the reference information storage means 107 showing the fourth embodiment of the present invention. The characteristic information of the reference information storage means 107 is a series of numbers in the order of detection of the characteristics of the wiring pattern extracted from the characteristic extraction means 106 using a non-defective substrate in advance, coordinates, a code by which the characteristics can be identified, and a series of a plurality of peripheral points. It is converted into a number format and registered. FIG. 8 shows an embodiment in which the number of characteristic points is four digits and the number of peripheral points is neighborhood 1 to neighborhood 4.

(Fifth Embodiment) The fifth embodiment of the present invention will be described below. FIG. 9A shows a fifth embodiment of the present invention and is a diagram showing region division when obtaining a plurality of peripheral feature points to be registered as one of the feature information of the reference information storage means 107. .

In FIG. 9A, the area is divided into four areas centering on the feature point of interest, and the closest feature point is selected in each area. The feature point 805 of interest is divided into regions 1 to 4, and the feature points closest to each region are the peripheral feature points 806 and 8.
It is registered as 07.

In addition, the area to be divided is not limited to the cross shape in FIG.
As shown in (b), the area is 45 ° with respect to the feature point of interest,
The closest feature points may be selected by dividing each of the X-shaped patterns of 135 °, 225 °, and 315 ° into four.

(Sixth Embodiment) The sixth embodiment of the present invention will be described below. FIG. 10A shows a configuration diagram of the feature point map of the comparison and determination means showing the sixth embodiment of the present invention, which will be described below.

In FIG. 10A, M × of the substrate to be inspected
The M inspection board entire area 900 is divided into grid-like divided areas 901 of an arbitrary size. The feature point map is shown in FIG.
As shown in the memory configuration of FIG. 5, each of the divided coordinates 905, which is a divided area, is composed of a registered characteristic point area 906 in which the detected characteristic point is registered in the area. The feature point map is obtained by comparing and determining means 109 with the feature point area 902 of interest.
And a reference region 903 including the target feature point region is set, for example, when selecting a peripheral feature point of the target feature point when converting into reference information data as shown in FIG. 8 or a peripheral feature point on the inspected substrate. When searching for, the feature points are searched at high speed.

(Embodiment 7) A seventh embodiment of the present invention will be described below. The seventh embodiment is an improvement of the sixth embodiment of the present invention, and is intended to reduce the memory capacity of the feature point map. This will be described below with reference to FIGS. 10, 11 and 12.

In order to reduce the memory capacity of the feature point map, first, the memory structure of the feature point map is shown in FIG.
As described above, only the first feature point in the area is registered. Second, as shown in FIG. 12, an item called map 1001 is added to the reference information data, and the feature points in the same divided area describe the serial number of the next feature point in the order of occurrence. As a result, as shown in FIG. 12, the feature point 0002 is registered in the feature map as the first feature point, and the search order 1002 is the feature point 0004 → feature point 0009 → feature point 0020 → in the same divided area. As shown, it is a chained search. As a result, the same effect as the sixth embodiment can be obtained, and the memory capacity of the feature point map can be reduced.

(Embodiment 8) The eighth embodiment of the present invention will be described below. FIG. 13 is a conceptual diagram showing the order of performing the inspection according to the eighth embodiment of the present invention, which will be described below.

When the comparison / determination means 109 compares and determines the feature points of interest, it does not determine the feature points of interest in the order of occurrence, but the peripheral feature referred to when the first feature point 1101 of interest as shown in FIG. 13 is compared and determined. Point next candidate feature point 1102
When the determination of the first feature point 1101 of interest is completed, it is selected as the feature point of interest from among the next candidate feature points 1102. Further, the peripheral feature points referenced by the next candidate feature point 1102 are successively selected as candidate feature points 1103. The peripheral feature points registered and referred to as are processed in a chained manner, and the determination process can be efficiently performed because the process is performed while using the related feature points.

If the processing is performed in the order of occurrence, it is wasteful to collect and generate data of peripheral characteristic points again and to process far characteristic points even though they are in the vicinity.

(Ninth Embodiment) The ninth embodiment of the present invention will be described below. FIG. 14 is a conceptual diagram for explaining the ninth embodiment of the present invention, which will be described below. In FIG.
The feature points of the reference information data are corrected with high accuracy by comparing the coordinates of the feature points which have already been associated with each other, and the characteristic points 1104 and 1106 of the already determined reference information data are compared. Feature point 1 of feature information data
The coordinate deviation of 107, 1105 is calculated by (Equation 1), and the correction amount of the coordinate is obtained to correct the coordinate of the characteristic point 1108 of the reference data.

[0069]

[Equation 1]

The correction processing is performed by the comparison / determination means 10
9 is performed sequentially to correct a positional error due to expansion / contraction of the printed circuit board 101, unevenness of etching, and the like, and perform an accurate inspection.

Further, although the calculation of the coordinate correction is described by the average value in (Equation 1), weighting or the like may be performed according to the minimum value or the maximum value or the distance between the target feature point and the candidate point.

(Tenth Embodiment) The tenth embodiment of the present invention will be described below. FIG. 15 is a block diagram showing the tenth embodiment, and FIG. 24 is a diagram specifically showing the processing, which will be described below. FIG. 15 is a block diagram showing a specific example of the feature extracting means 106.

First, the feature extraction means 106 inputs the binary image 1203 from the binarization means 105, performs thinning processing 1201 for removing each pixel from the outside of the wiring pattern while maintaining connectivity, and obtains a skeleton image, Feature extraction process 12
At 02, characteristic information data 1204 such as an end point and a T-branch is obtained. The obtained characteristic information data 1204 is processed by the comparison / determination unit 109 prior to the comparison / determination processing, and when another characteristic point exists in the area near the characteristic point, the characteristic information data is rewritten by a combination of characteristic codes described later. Is stored in the characteristic information storage means 108. Although not shown in FIG. 15, when the reference information data is created, the characteristic information data is rewritten and stored in the reference information storage means.

Next, the feature extraction means 106 and the comparison determination means 109 will be described in detail. Feature extraction means 10
6 includes a thinning process 1201 and a feature extraction process 1202, which will be described with reference to FIGS. 16 to 21. FIG. 16 is a detailed block diagram of the thinning process.
In order to process the thinning processing unit which receives the binary image 1205 from the quantizing means 105 and cuts one pixel from the outside of the wiring pattern while maintaining the connectivity in real time, the unit # 1.
(1206) to unit #n (1209) are connected to the n-stage pipeline to output a skeleton image 1211. The thinning processing is an image processing well known as a method of line drawing processing. For example, “Tamura:“ Considerations on the thinning method ”. PRL 75-66 (1975) ) ”. Further, Japanese Patent Laid-Open No. 04-7375 described in the conventional example.
Since it is also described in detail in Japanese Patent Publication No. 9, the description is omitted. However, in general, the thinning process is performed by cutting each pixel from the outside of the pattern until a core line having a width of one pixel is completely formed, but here, from the viewpoint of real-time processing and economical point, It performs a finite number of times.

The feature extraction processing 1202 is the thinning processing 12
The skeleton image from 01 is input and the features of the end points and the T-branch are extracted from the skeleton image. FIG. 17 shows a detailed block diagram of the feature extraction processing, which will be described below. Figure 1
7, the skeleton image 1215 from the thinning processing 1201 is input and the line memory 1221 and the scanning window 121 are input.
6 and LUT1 (look-up table) 1217 to perform 3 × 3 mask processing. LUT1 (1217)
Data for detecting patterns as shown in FIG. 19, FIG. 20 and FIG.
8 and T branch detection 1219 and cross flag detection signals are output. Further, based on the synchronization signal 1224, a coordinate generator 1225 outputs a feature identification code such as an end point or T-branch and a coordinate 1226 as feature information data 1227.

FIG. 18 shows an embodiment of the format of characteristic information data consisting of coordinates (X: main scanning, Y: sub scanning) and a characteristic identification code. The cross flag is used to distinguish the cross branch from the T branch as described later, and all the branches are detected as T branches, but the cross flag is detected only when the cross branch is taken. Next, the comparison / determination unit 109 will be described, and its processing flow is shown in FIG. This process is performed prior to the comparison / determination process, and when another feature point exists in the area near the feature point, the feature information data is rewritten by a combination of feature codes described later and stored in the feature information storage means 108. This will be described below.

In process i, the characteristic information from the defect extracting means is input and temporarily stored. The process j checks whether or not there are other feature points in an arbitrary neighborhood area of the feature point of interest. The process k is
If not, it is stored in the feature information storage means as it is. Processing l
If there are other characteristic points, the characteristic information is rewritten by a combination of characteristic codes and then stored in the characteristic information storage means.

Incidentally, the processing a to processing h is for carrying out the comparison judgment processing according to the flow shown in FIG. 5 or FIG. An example of the combination of feature identification codes and how to rewrite them will be described with reference to FIGS.
However, in the figure, the skeleton image is described as being rewritten for the purpose of explaining the concept, but the feature information data is actually rewritten.

For example, in the first embodiment shown in FIG. 24, one T-branch (three-branch) and one end point are detected in the neighborhood area 1302, and it is deleted because there is no feature point. This is because the beard is generated due to the shape of the edge and is not a defect. Also, FIG. 25 and FIG.
The same applies to FIG. 28. In the case of FIG. 29, depending on the edge shape of the wiring pattern, a + character (4 branches) or two T branches (3 branches) are detected or become unstable.
When four branches are detected, it is rewritten by dividing into three branches × 2.

Although the present embodiment has been described with reference to the embodiment in which the processing is performed prior to the comparison / determination processing, even if the identification code of the characteristic information data is rewritten before being registered in the error information memory 112 in the processing f. The same effect can be obtained.

(Eleventh Embodiment) The eleventh embodiment of the present invention will be described below. FIG. 15 is a block diagram showing an eleventh embodiment of the present invention, and FIG. 30 is a diagram showing a concrete state of the processing, which will be described below.

First, FIG. 15 is a block diagram showing a concrete embodiment of the defect detecting means 106 as described above.

The defect detecting means 106 is a binarizing means 105.
The thinning process 12 in which the binary image 1203 from is input and the pixel is removed from the outside of the wiring pattern pixel by pixel while maintaining connectivity.
01 is obtained to obtain a skeleton image, and the boundary flag is simultaneously output from the nkth stage and thereafter of the thinning process. In the feature extraction processing 1202, feature information data 1204 such as end points, T-junctions, and boundary flags are obtained. The obtained characteristic information data 1204 is subjected to the comparison / determination processing by the comparison / determination means 109, and when the error is detected, the boundary flag is checked, and if it is detected, it is excluded from the error. FIG. 16 shows a detailed block diagram of the thinning processing. The thinning processing unit for inputting the binary image 1205 from the binarizing means 105 and shaving one pixel while maintaining connectivity from the outside of the wiring pattern is shown in real time. N of unit # 1 (1206) to unit #n (1209) for processing
It is connected to a stage pipeline and outputs a skeleton image 1211. At the same time, the thinning processing unit #
The skeleton image 1213 is branched from nk (1208) and output via the delay memory 1210 for adjusting the timing with the skeleton image of the final-stage thinning processing unit #n. FIG. 17 shows a detailed block diagram of the feature extraction processing, in which the skeleton image 1215 from the thinning processing 1201 is input and the line memory 1221 and the scanning window 1 are input.
216 and LUT (Look Up Table) 1217
Then, 3 × 3 mask processing is performed to extract features such as the end point detection 1218, the T branch detection 1219, and the detection signal 1228 which is a logical sum signal of the end point detection and the T branch detection. Further, the delayed skeleton image 1221 is input, and the line memory 1222, the scanning window 1229, and the LUT2 (1
The mask processing is performed according to 230). LUT2 here
(1230) is the contents of the scan window 1229 and LUT1 (1
The boundary flag detection 1220 is output by the detection signal 1228 from 217). LUT2 (1230)
Is a pixel width from the delayed skeleton image 1221 from the thinning process # n-k and LUT1.
Based on the presence / absence of the detection signal 1228 from (1217), if there is a detection signal with one pixel width, the boundary flag detection 1220 is "
If there is a detection signal instead of one pixel width, the boundary flag detection 1220 is set to "1". End point detection 1218 and T
Coordinates 1226 generated by the coordinate generator 1225 based on the synchronization signal 1224 are output as feature information data 1227 together with the feature identification codes such as the branch detection 1219 and the boundary flag detection 1220. Next, the comparison and determination means 109
FIG. 23 shows the processing flow thereof.

In FIG. 23, the comparison / determination processing of processing a to processing e and processing g to processing h is the same as that of the first embodiment, and therefore its explanation is omitted. Only the processing of f different from (Example 1) will be described below.

First, a defect is detected in the comparison and judgment processing of a to e, and if the boundary flag is added to the characteristic information data which is the processing of f, the characteristic information data is not registered as a defect and the boundary flag is The feature information data that has not been added is registered in the error information storage memory 112 as an error.

A concrete example is shown in FIG. 30 and will be described.
14 (a) and 14 (b) show an example of the end point. In the case where the skeleton image is completely one pixel wide due to the variation in the manufacturing process of the printed circuit board by the finite number of thinning processes.
When it does not become 01, 1402 occurs.

Also in the case of FIGS. 30 (c) and 30 (d),
There is a case where the pixel width is completely one pixel and it is detected as a T branch, and a case where the pixel width is not one pixel and the T branch is not detected, which is extremely unstable as an inspection device. Therefore, in the finite number of times of thinning processing, a false flag is suppressed by adding a boundary flag to a feature point near a boundary condition with a width of one pixel and excluding it from the defect data when comparing and judging. be able to.

[0088]

As described above, the effects of the present invention are
First, in the comparison / determination means, it is determined whether the feature point rn from the non-defective substrate corresponds to an arbitrarily determined first determination region centered on the feature point tn of interest extracted from the inspected substrate in the first determination process. The distance sl between the target feature point tn and the one or more feature points tn-i that have already been associated with each other is determined in the second determination process, and the peripheral feature points rn- of the non-defective substrate that have already been associated with each other. By determining whether or not there is a feature point rn corresponding to an arbitrary second determination region centered on a point with a distance sl from i, a plurality of candidate points can be surely associated and a highly reliable inspection device can be provided. It is possible to provide it, and the effect is great.

Secondly, in the second determination processing of the comparison / determination means, the distance between the feature point of interest and one or more peripheral feature points is obtained, and an arbitrary distance is set around the point of distance from each candidate point of the non-defective substrate. It is determined whether or not there are feature points in the second determination area, a score is given to each existing feature point, and the candidate point with the highest score is selected as the corresponding feature point. It is possible to provide a highly reliable inspection device capable of reliably associating with each other.

Thirdly, if it cannot be specified in the first judgment processing and the second judgment processing of the comparison judgment means, the third judgment processing is performed.
As a determination process of, the distance between the feature point of interest and one or more peripheral feature points is obtained, and it is determined whether the feature point exists in an arbitrary third determination area centered on a point having a distance sl from the candidate point of the non-defective substrate. A highly reliable inspection device capable of reliably associating a plurality of candidate points is provided by judging and giving a score for each existing feature point and selecting the candidate point with the highest score as the corresponding feature point. It becomes possible to provide.

Fourth, since the wiring pattern feature information is made up of a series of numbers in the order of detection, coordinates, a code capable of identifying the feature, and a sequence number of a plurality of surrounding feature points, the peripheral points can be easily searched. Therefore, it is possible to provide an efficient and high-speed inspection device, and the effect is great.

Fifth, when obtaining a plurality of peripheral feature points to be registered as one of the feature information of the reference information storage means, the region is divided into four regions with respect to the feature point of interest, and the feature point closest to each of them is obtained. It is possible to provide a highly reliable inspection device because the peripheral feature points are not concentrated in one place, and the effect is great. In addition, the area is 45 ° or 135 ° with respect to the feature point of interest.
Multiple CCs with 225, 325, and 315 ° divisions and feature points with the closest distances from each are selected.
This is for avoiding dispersion of peripheral feature points when a region is divided by the D camera and inspected, and it is possible to provide a highly reliable inspection device, which has a great effect.

Sixth, in the comparing and judging means, the characteristic information from the characteristic extracting means is registered in the corresponding area in the characteristic point map obtained by dividing the overall coordinates of the printed circuit board into arbitrary sizes, and the peripheral characteristic of the characteristic point of interest. Since points are searched from the feature point map, it is possible to provide a high-speed inspection device because it is possible to search for surrounding feature points at a high speed without the need of brute force search. There is something.

Seventh, in the comparison / determination means, the first feature information from the feature extraction means is registered in the corresponding area in the feature point map obtained by dividing the overall coordinates of the printed circuit board into arbitrary sizes, and in the divided area. For the existing second and subsequent feature points, a serial number is added to the previously detected feature information, and the peripheral feature points of the target feature point are searched in a chain from the feature point map and the serial number added to the feature information. By doing so, the memory size of the feature point map is reduced, and it is possible to provide a small-sized and high-speed inspection apparatus, and the effect is great.

Eighthly, when the comparison / determination means compares and determines the feature point of interest from the feature information storage means, the peripheral feature points are registered as the next search candidates, and after the comparison determination of the feature points of interest, the next search candidate is selected. It is possible to provide a high-speed and highly reliable inspection device because it is possible to process efficiently because the next feature points of interest are searched in a chained manner and the processes are processed in order from those with strong relevance. The effect is great.

Ninth, in the comparison and determination means, the coordinates of the feature points from the reference information storage means are sequentially corrected by the average value of the deviation amounts of the one or more feature points that have already been determined. It is easy and absorbs the variation in the process, and it is possible to provide an inexpensive and highly reliable inspection device, and the effect is great.

Tenth, in the feature extracting means, when converting a binary image into a skeleton image by thinning processing or the like and extracting a feature of a wiring pattern from an end point or a branch point, a feature in an arbitrary neighborhood region of the feature point The identification code of the feature is re-attached according to the combination of points, and it provides a highly reliable inspection device that absorbs instability due to variations in the etching process and does not cause false alarms, and its effect is great. .

Eleventh, in the feature extracting means, when the binary image is converted into a skeleton image by a finite number of times of thinning processing of n times and the features of the wiring pattern are extracted from the end points or the branch points, the n-th time A flag is added to the feature point in the subsequent thinning process and is excluded when an error is determined in the comparison determination. This is a highly reliable inspection device without false alarm that absorbs instability due to variations in etching process. And the effect is great.

[Brief description of drawings]

FIG. 1 is a basic block connection diagram of a wiring pattern inspection apparatus according to a first embodiment of the present invention.

FIG. 2 is a conceptual diagram explaining a first determination process in the first embodiment.

FIG. 3 is a conceptual diagram illustrating a second determination process in the first and second embodiments.

FIG. 4 is a conceptual diagram illustrating a third determination process in the third embodiment.

FIG. 5 is a processing flow chart of a comparison / determination unit in the first embodiment.

FIG. 6 is a processing flow chart of a comparison / determination unit in the second embodiment.

FIG. 7 is a processing flow chart of a comparison / determination unit in the third embodiment.

FIG. 8 is a conceptual diagram showing a format of characteristic information of a reference information storage means, which is an essential part of the third embodiment.

FIG. 9 is a conceptual diagram showing area division of peripheral feature points in the fourth and fifth embodiments.

FIG. 10 is a conceptual diagram of a feature point map of comparison and determination means in the sixth embodiment.

FIG. 11 is a memory conceptual configuration diagram of feature point maps of the comparison and determination means in the sixth and seventh embodiments.

FIG. 12 is a diagram showing a chained search of a feature point map of the comparison determination means in the seventh embodiment.

FIG. 13 is a conceptual diagram showing the order in which the inspection of the comparison determination means in the eighth embodiment is performed.

FIG. 14 is a conceptual diagram showing correction of coordinates of a comparison and determination means in the ninth embodiment.

FIG. 15 is a block connection diagram of a defect extracting means in the tenth embodiment.

FIG. 16 is a detailed block connection diagram of thinning processing in the tenth embodiment.

FIG. 17 is a detailed block connection diagram of feature extraction processing in the tenth embodiment.

FIG. 18 is a diagram showing a format of characteristic information data in the tenth embodiment.

FIG. 19 is an end point detection pattern diagram of feature extraction processing in the tenth embodiment.

FIG. 20 is a flowchart of the feature extraction process T in the tenth embodiment.
Branch detection pattern diagram

FIG. 21 is a cross-branch detection pattern diagram of the feature extraction processing in the tenth embodiment.

FIG. 22 is a process flow chart of the comparison / determination means in the tenth embodiment.

FIG. 23 is a process flow chart of the comparison / determination means in the eleventh embodiment.

FIG. 24 is a conceptual diagram of rewriting a feature identification code in the tenth embodiment.

FIG. 25 is a conceptual diagram of rewriting the feature identification code in the tenth embodiment.

FIG. 26 is a conceptual diagram of rewriting the feature identification code in the tenth embodiment.

FIG. 27 is a conceptual diagram of rewriting the feature identification code in the tenth embodiment.

FIG. 28 is a conceptual diagram of rewriting the feature identification code in the tenth embodiment.

FIG. 29 is a conceptual diagram of rewriting the feature identification code according to the tenth embodiment.

FIG. 30 is a specific conceptual diagram for a finite number of thinnings in the eleventh embodiment.

FIG. 31 is a process flow chart of a conventional wiring pattern inspection device.

[Explanation of symbols]

 101 Printed Circuit Board 102 CCD Camera 103 Image Input Means 104 Illumination Means 105 Binarization Means 106 Feature Extraction Means 107 Reference Information Storage Means 108 Feature Information Storage Means 109 Comparison Determination Means 110 CPU System 112 Error Information Storage Memory 201 Feature Information Coordinate System 202 attention feature point 205 1st determination area 206 peripheral feature point 302 attention feature point 303 feature point with completed association 307 2nd determination area 801 to 804 division area 808 to 811 division area 901 division area 902 attention feature point area 903 Reference area 904 Division coordinate 905 Division coordinate 906 Registered feature point area 1001 Map 1002 Search order 1101 Interesting feature point 1102 Next candidate feature point 1103 Next candidate feature point 1201 Thinning process 1202 Feature extraction process 12 3 Binary image 1204 Feature information data 1206 to 1209 Thinning processing unit 1210 Delay memory 1211 Skeleton image 1216 Scan window 1217 LUT 1225 Coordinate generator 1300 Wiring pattern 1301 Skeleton 1302 Neighboring area 1303 to 1307 Target feature point 1400 Wiring pattern 1402 Skeleton 1401 Skeleton ~ 1403 Target feature points

Continuation of front page (51) Int.Cl. ⁶ Identification number Office reference number FI Technical display location G06T 7/00 H01L 21/66 J 7630-4M H05K 3/00 Q

Claims (11)

[Claims] 1. An image input unit for photoelectrically converting a wiring pattern formed on a printed circuit board, a binarizing unit for converting a grayscale image from the image input unit into a binary image, and the binarizing unit. Feature extracting means for extracting features of the wiring pattern from the binary image, and feature information storing means for temporarily storing the feature information of the wiring pattern from the feature extracting means,
Reference information storage means for extracting characteristic information of the wiring pattern from the characteristic extraction means in advance from a non-defective board and storing it as reference data; characteristic information of the wiring pattern of the board to be inspected and the reference information storage from the characteristic information storage means. The characteristic point rn from the non-defective substrate corresponds to an arbitrarily determined first determination area centered on the characteristic point tn of interest extracted from the inspected substrate as the first determination processing based on the characteristic information of the non-defective substrate from the means. If there is no corresponding feature point, it is registered as a defect, and if there are a plurality of corresponding feature points, one or more already associated with the feature point tn of interest as the second determination process. Distance sl from the surrounding feature point tn-i of
And the target feature point tn is associated with the second determination region arbitrarily determined around the point of distance sl from the peripheral feature point rn-i that has already been associated with the peripheral feature point tn-i of the non-defective substrate. The wiring pattern inspection apparatus is characterized by comprising: a comparison / determination unit that determines whether or not the characteristic point rn is present, and if not, registers it as a defect and detects only a true defect. 2. The comparison / determination means sets a feature point rn from a non-defective substrate in an arbitrarily determined first determination region centered on a feature point tn of interest extracted from the inspected substrate as the first determination process.
Correspond to each other. If there is no corresponding feature point, it is registered as a defect. If there are a plurality of corresponding candidate points, the target feature point tn and one or more peripheral feature points are used as the second determination process. Is calculated, it is determined for each candidate point of the non-defective substrate whether or not the peripheral feature point exists in the second determination area arbitrarily set around the point of the distance from the peripheral feature point, and the feature point exists. 2. The wiring pattern inspection apparatus according to claim 1, wherein a score is given for each and a candidate point having the highest score is selected as a corresponding feature point rn. 3. The comparison / determination means sets a feature point rn from a non-defective substrate in an arbitrarily determined first determination region centered on a feature point tn of interest extracted from the inspected substrate as a first determination process.
Correspond to each other. If there is no corresponding feature point, it is registered as a defect. If there are a plurality of corresponding feature points, as the second determination process, the target feature point tn has already been associated with 1 A distance sl to one or more peripheral feature points tn-i is obtained, and the distance sl is set to a second determination region that is arbitrarily set around a point of a distance sl from the peripheral feature point rn-i of the non-defective substrate that has already been associated. It is determined whether or not the corresponding feature point rn exists, and if it does not exist, it is registered as a defect, and if there are a plurality of corresponding candidate points and it cannot be specified, then as the third determination process, the feature point tn of interest and one or more surroundings. The distance to the feature point is obtained, and it is determined whether or not the peripheral feature point exists in a third determination area that is arbitrarily set around the point of the distance from the peripheral feature point for each candidate point of the non-defective board. Score is given each time there is Wiring pattern inspecting apparatus according to claim 1, wherein any one, characterized in that selected as feature points rn corresponding auxiliary points. 4. The reference information storage means comprises, as the characteristic information of the wiring pattern from the characteristic extracting means, a series of numbers in the order of detection, coordinates, a code capable of identifying the characteristic, and a series of a plurality of peripheral characteristic points. Claims 1, 2, 3 characterized
Any one of the wiring pattern inspection devices. 5. When obtaining a plurality of peripheral feature points to be registered as one of the feature information of the reference information storage means, the region is divided into four around the feature point of interest and the feature of the closest distance in each region. The wiring pattern inspection apparatus according to claim 4, wherein points are selected. 6. The comparison / determination means sets a feature point map obtained by dividing the entire area of the printed circuit board into arbitrary sizes, registers the feature information from the feature extraction means in the corresponding area, and surrounds the feature point of interest. The wiring pattern inspection apparatus according to any one of claims 1, 2, 3, 4, and 5, wherein characteristic points are searched from a characteristic point map. 7. The comparison / determination means registers the first feature information from the feature extraction means in a corresponding area in a feature point map obtained by dividing the entire area of the printed circuit board into an arbitrary size, and within the divided area. When registering the second and subsequent feature points existing in, the serial number of the feature point is added as map information to the previously detected feature information of the reference information storage means, and the peripheral feature points of the feature point of interest are added. 7. The wiring pattern inspection device according to claim 6, wherein the feature point map and the map information added to the feature information of the reference information storage means are searched in a chained manner. 8. The comparison determination means registers peripheral feature points as next search candidates when comparing and determining the feature points of interest from the feature storage means, and after completion of the comparison determination of the feature points of interest, the next search candidate is selected. The wiring pattern inspection apparatus according to any one of claims 1, 2, 3, 4, 5, 6, and 7, wherein the characteristic points are searched in a chained manner. 9. The comparison determination means sequentially corrects the coordinates of the feature points from the reference information storage means with the average value of the deviation amounts of the one or more feature points that have already been determined. The wiring pattern inspection device according to any one of 2 and 3. 10. The comparing and determining means converts the binary image from the binary image by the feature extracting means into a skeleton image by a thinning process or the like to extract the features of the wiring pattern such as the end points and the branch points, and selects any of the feature points at the comparison and determination. 10. The wiring pattern inspection device according to claim 1, wherein the feature identification code is reattached in accordance with a combination of feature points in the neighborhood area. . 11. The comparison and determination means converts the binary image from the binary image into a skeleton image by a finite number of thinning processing of n times or the like by the feature extraction means and extracts the features of the wiring pattern such as end points and branch points, 11. The wiring pattern inspection apparatus according to claim 10, wherein a flag is added to the feature points during the nk-th thinning process and thereafter, and the feature points are excluded when an error is determined during the comparison determination.